QMAK

QMAK is a GUI tool for explainable machine learning allowing users to interact with the trained models and visualizing classification process. It helps to:

  • understand why a particular decision was selected for a classified object
  • find out how much a classifier is sure of the assigned decision
  • find out which element of a classifier needs to be improved and modify it interactively
This one visualization tool aggregates multiple models and allows the community to add more visualized models through a simple connecting interface. It uses Rseslib library as the source of classification models.

Features

  • Visualization of data, classifiers and single object classification
  • Interactive classifier modification by a user
  • Classification of test data with presentation of misclassified objects
  • Experiments comparing classification accuracy of classifiers with various test types
  • Easy addition of classifiers implemented by users
Download

QMAK is part of Rseslib. To run QMAK Java 8 or later needs to be installed in the system.
To start QMAK unpack the rseslib-3.x.x.zip package and run the script qmak.sh (on Linux) or qmak.bat (on Windows).
Help on QMAK can be found in the main menu of the application.
FedCSIS 2023 paper provides detailed description of QMAK.
All classifiers available in QMAK are described in Chapter 13 Classifier types of Rseslib User Guide.

Publication

Wojna, A., Jachim, K., Kosson, L., Kowalski, L., Manski, D., Manski, M., Mroczek, K., Niemkiewicz, K., Piszczatowski, R., Prochniak, M., Romanczuk, T., Skibinski, P., Staszczyk, M., Szostakiewicz, M., Tur, L., Wojcik, D., Zuchniak, M., QMAK: Interacting with Machine Learning Models and Visualizing Classification Process, FedCSIS 2023, In Proceedings of the 18th Conference on Computer Science and Intelligence Systems, 2023, pp. 315-318, IEEE, [PDF]